I'm copying in today's update from Stratechery on Intel. It is obviously Intel focussed - and they are ina lot of trouble - but there are lot's of great insights about what is happening in the industry and why.

Intel is the 8th largest holding in SEMI at 4.6% of the assets.

DISC: I hold a small amount of SEMI in my SMSF

Intel Earnings

From Bloomberg:

Intel Corp. is forecasting one of the worst quarters in its history, touching off a broader selloff of chips companies as a slowdown in personal-computer sales ravages the industry. Late Thursday, Intel gave a sales range that missed analysts’ estimates by billions of dollars, warning that revenue could fall to the lowest quarterly level since 2010. Shares plunged 10%, the most since July, in New York. Advanced Micro Devices Inc., Applied Materials Inc., Lam Research Corp. and Qualcomm Inc. declined. Semiconductor maker KLA Corp., which issued its own disappointing outlook, also slid.

Chip companies are reeling over a steep decline in demand for PC processors that has wiped out profits and led to deep cuts across the industry. Intel’s outlook signals more pain to come. The company is eliminating jobs and slowing spending on new plants in an effort to save as much as $10 billion. It’s taking an especially large hit from losing market share to rivals…

It was a painful admission for a company that has been attempting a multiyear comeback under Chief Executive Officer Pat Gelsinger, who took the helm in 2021. A post-pandemic downturn for Intel’s main business, PC chips, has torpedoed efforts to get the company’s financial performance back on course. Instead, it’s only losing more ground. “I’d like to remind everyone that we’re on a multiyear journey,” Gelsinger said during a conference call.

For almost the entire run of Stratechery I have been warning that Intel was in trouble; during that run Intel reported profitable quarter after profitable quarter, with the stock marching up-and-to-the-right.

The biggest driver was exploding usage in the data center; yes, Intel missed out on mobile, but mobile computing is intertwined with cloud computing, and Intel was the chief beneficiary of the parallel growth in the latter.

The trouble with missing mobile was threefold: first, Intel missed out on the massive volumes that would, in the long run, be important for paying off ever more expensive fabs; instead those volumes went to TSMC. Second, TSMC, fueled by the massive revenue influx that came from delivering those volumes, accelerated its march down the process curve, beating Intel by years to the very expensive and very risky EUV transition that allowed TSMC to leap ahead in performance. Third, mobile was, by necessity, at the forefront of prioritizing efficiency and performance-per-watt given the limitations of mobile devices. Today, though, efficiency and power-per-watt increasingly matter for servers, as well.

The net result is that today it is AMD, the second supplier IBM insisted on way back at the dawn of the PC era, that makes the best data center chips thanks in part to TSMC’s process advantage, even as data center demand has slowed dramatically in the inevitable-yet-late-to-arrive semiconductor downturn we are currently experiencing, and as data center operators like AWS invest in their own ARM-based chips; Intel, meanwhile, still provides a lot of volume in PCs — thanks in part to very aggressive discounting — but the pandemic pulled forward huge amounts of PC buying: that worked to Intel’s benefit in 2020 and 2021, but now the hangover is hitting and it is a doozy.

Intel’s Plunging Margin

This has resulted in a triple-whammy to Intel’s earnings:

• First, volumes were down dramatically thanks to high inventories amongst OEMs (a less charitable interpretation is that Intel stuffed the channel).
• Second, prices are down thanks to Intel needing to increasingly discount to counter AMD’s total-cost-of-ownership advantage.
• Third, the largest cost of goods sold for a chip is fixed costs, the depreciation of which is incurred no matter how many chips are sold; that means that fewer chips sold for lower prices has an outsized impact on gross margins.

Indeed, Intel’s gross margins were a shocking-for-Intel 39% (notice how revenue declined much more than COGS):

It gets worse, though: first, Intel forecast a gross margin for next quarter of 34%. Second, Intel’s margins next quarter will include a depreciation adjustment for some equipment from 5 years to 8 years; that will decrease COGS in the above chart, and increases gross margin, which means that 34% forecast is actually higher than it would be! Intel explained on its earnings call:

Before turning to Q1 guidance, let me take a moment to discuss an accounting change that will impact our results beginning in the first quarter. Effective January 2023, we increased the estimated useful life of certain production machinery and equipment from five years to eight years. This change better reflects the demonstrated economic value of our machinery and equipment over time and is more aligned with the business model changes inherent to our IDM 2.0 strategy.

The growth of the IFS deal pipeline will extend the life of manufacturing nodes beyond what was practical within IDM 1.0. Disaggregated CPU architecture allows performance and cost optimization for each chiplet better leveraging older nodes. And we are optimizing our core business around more sustainable capacity quarters to improve equipment utilization and maximize ROIC.

The change will be applied prospectively beginning Q1 2023. When compared to the estimated useful life in place as of the end of 2022, we expect total depreciation expense in 2023 to reduce by roughly $4.2 billion. An approximate $2.6 billion increase to gross profit, a $400 million decrease in R&D expense and a $1.2 billion decrease in ending inventory values.

This change is, on its face, a reasonable one: using expensive equipment for longer is a core part of the economics undergirding the foundry business, and has long been an important part of TSMC’s profitability. Still, it’s worth noting that TSMC depreciates its equipment over five years; that casts Intel’s change, even if justified, in a significantly less flattering light.

The Dividend Question

Still, at the end of the day depreciation schedules are about numbers on an income statement; what is actually more important, at least for Intel these days, is cash flow, and it’s not pretty: the company’s adjusted free cash flow, which only considers operating activities and Intel’s investments in property, plant, and equipment, was negative $4 billion. Intel’s overall cash flow, including equity sales, the MobileEye IPO, and debt issuances was still $15 billion in the black, but it’s the adjusted number that is a better signifier of Intel’s long-term prospects.

The problem Intel faces is not simply its declining revenue: that declining revenue is downstream from investments and advancements Intel didn’t make for many years; that’s why I opened this Update recounting the concerns I have had for a very long time. In this case, I don’t think I was wrong to raise those concerns: rather, it’s an example of how the semiconductor industry, thanks to its massive lead times, massive capital requirements, and massive uncertainty, takes many years for the consequences of decisions and investments to bear fruit. In the case of Intel today, that fruit is rotten: CEO Pat Gelsinger is taking the heat for Intel’s terrible results, but these results are downstream from the CEOs that preceded him. The inverse of that reality, though, is that bearing good fruit will also take time. Gelsinger said in his prepared remarks:

I want to remind everyone that, we are on a multi-year journey. We remain focused on the things that are within our control as we navigate short-term headwinds, while executing to our long-term strategy. While I remain sober that, we have a long way to achieve our financial expectations, I am pleased with the transformation progress that we are making.

Here is the fundamental challenge Gelsinger faces: all of the CEOs that led Intel into the current ditch — from Paul Otellini to Brian Krzanich (the two most guilty parties) to Bob Swan — mostly got to preside over happy earnings calls, the results of which were due to investments and decisions made by Robert Noyce, Gordon Moore, and Andy Grove (Craig Barrett was more of a mixed bag); the problem is that Intel’s investors — as evidenced by Intel’s rising stock price over the last decade, even as the rot under the surface became more and more apparent — didn’t care about the decade-long interval between decisions and their consequences. And, to be fair, I can understand the investor point-of-view: had you avoided investing in Intel according to my expressions of concern you would have missed out on many years of gains.

The implication of this lack of understanding or caring for the long-lead time for Intel decision-making and investments is that Gelsinger is probably not getting the benefit of the doubt for the investments and direction Intel is heading now, including the IDM 2.0 strategy and the commitment to advance five nodes in four years. A lot of that skepticism is justified, to be clear: Intel’s foundry strategy is very risky, as the company needs to not only catch up to TSMC in terms of process but also create a customer service organizationfrom scratch, which entails completely reworking its culture. And, as noted, Intel needs to catch up to TSMC, which is a tough sale given that Intel spent the previous decade falling behind the Taiwanese foundry. No one is going to believe that Intel can do it until they do, and that includes investors.

Indeed, the smartest play for investors is probably to wait for evidence that Intel is succeeding in its turnaround, and only then try and capture the upside — the inverse of investors who held Intel stock and profited in the very decade when the company was falling apart. Again, though, that is several years out, and Gelsinger needs investor support now. I think he is on the right course, but that course is both uncertain to succeed and even if it does, won’t bear fruit for a while.

This, I think, explains why Intel is still paying a dividend, a point of consternation for many semiconductor analysts. I get their objection: Intel needs cash if it is going to pull this transformation off, so why is it paying investors a dividend for the 122nd quarter in a row?

I am sure the answer has a lot to do with that longevity: Intel is not technically a “dividend aristocrat”, which requires the company to increase its dividend every year for 25 years; the company froze its dividend in 2014, and, it’s worth pointing out, stopped share repurchases. Still, Intel has been a dividend-paying stock for over 30 years, and that’s meaningful to a certain set of investors; it’s also, per the previous point, the most compelling reason to own the stock for the next year or two.

That’s not nothing: not only is Intel owned by investors — at the end of the day, dividends are their money — but Gelsinger ultimately is employed by them, and is asking a lot of them. Intel could have spun off its fabs and followed AMD to being another fabless chip designer dependent on TSMC; Gelsinger went the opposite way and is risking the company to build up a U.S. foundry alternative to TSMC. It was a rational decision — the worst thingIntel could have done is try to maintain the status quo of being an integrated device manufacturer that only made its own chips — but it will take a long time to pay off, and I think it’s reasonable for investors to want and receive a return in the meantime.

That’s not to say that Intel might not trim the dividend a bit; CFO David Zinsner said on the call:

We announced a $0.365 dividend for the first quarter. That was consistent with the last quarter’s dividend. I’d just say the Board, management, we take a very disciplined approach to the capital allocation strategy, and we’re going to remain committed to being very prudent around how we allocate capital for the owners. And we are committed to maintaining a competitive dividend.

Right now Intel’s dividend is not competitive, in that it is really high, thanks to Intel’s plummeting stock price; indeed, this upcoming dividend represents a dividend yield of around 6%, which is one of the highest in the entire stock market. Intel may very well give it a haircut, but I don’t think it goes away, and I think that’s ok. Gelsinger needs time, and if this buys him a bit more of it it will be a price worth paying.

This Update will be available as a podcast later today. To receive it in your podcast player, visit Stratechery.

The Stratechery Update is intended for a single recipient, but occasional forwarding is totally fine! If you would like to order multiple subscriptions for your team with a group discount (minimum 5), please contact me directly.

A cyclical industry indeed stays cyclical - Memory-chip sector suffering one of its worst routs ever | Fortune

"The memory-chip sector, famous for its boom-and-bust cycles, had changed its ways. A combination of more disciplined management and new markets for its products — including 5G technology and cloud services — would ensure that companies delivered more predictable earnings.

And yet, less than a year after memory companies made such pronouncements, the $160 billion industry is suffering one of its worst routs ever. There’s a glut of the chips sitting in warehouses, customers are cutting orders, and product prices have plunged."

I'd recommend a reading of Chip Wars by Chris Miller, I think it might even make FTs business book of the year.

The first half is a history of the semi conductor industry which I found interesting but some may find a little boring but explains how we ended up where we are with pretty much ASML and TMSC dominating their various niches and the differnt architectures such as how ARM plays into it all. Touches on Intel, Samsung etc as well. Also a bit on the military uses.

The second half goes into the currenty industry/politcal dynamics. Takeaway is just how fragile the supply chain in is and how reliant the world is on Taiwan with TMSC. If China tried to take Taiwan good luck getting a new car.

The massive news this week by the US to ban some chips from China is big news - What do US curbs on selling microchips to China mean for the global economy? | China | The Guardian. It will be extremley hard for China to completly internalise the whole chain - the sheer amount of trillions it would take the author posits puts it even out of reach of the Chinese Government.

First few lines from an article in the New York Times, worth logging into the NYT and having a read. Have to think it's a positive for SEMI

WASHINGTON — President Biden on Tuesday signed into law a sprawling $280 billion bill aimed at bolstering American chip manufacturing to address global supply chain issues and counter the rising influence of China, part of a renewed effort by the White House to galvanize its base around a recent slate of legislative victories.

Standing before business leaders and lawmakers in the Rose Garden, Mr. Biden said the bill was proof that bipartisanship in Washington could produce legislation that would build up a technology sector, lure semiconductor manufacturing back to the United States and eventually create thousands of new American jobs.

“Fundamental change is taking place today, politically, economically and technologically,” Mr. Biden said. “Change that can either strengthen our sense of control and security, of dignity and pride in our lives and our nation, or change that weakens us.”

latest from Economist on semiconductor industry:

Taiwan Semiconductor Manufacturing Company, which makes 90% of the world’s chips with the smallest, most efficient circuits, will report its quarterly results on Thursday. Investors expect a strong showing. Nonetheless, chipmakers face tougher times ahead. One problem is supply. Sales of laptops and smartphones, where half of chips are used, soared during covid-19 lockdowns. To meet demand, investment in production boomed. But much of the new capacity will arrive later this year, when demand is expected to drop. Another worry is geopolitics. Policymakers want to bring more of the chip manufacturing supply chain within their own borders. Such fragmentation would drive up costs and leave chip firms beholden to the whims of party politics. Another chip giant, Intel, has threatened to postpone the launch of two American factories because of a delay by Congress in passing a subsidies bill worth $52bn over five years to American chipmakers. A soaring TSMC may be about to fall back down to earth.

———————————————————————

Here is a link to a Morningstar article on the risks and benefits of investing in thematic ETFs.

https://www.morningstar.com.au/etfs/article/3-things-to-know-about-thematic-funds/201852

This point particularly resonated with me:

3. The odds are stacked against investors in thematic funds, but the prospective payouts are big

Investors in thematic funds are making a three-way bet. They are betting that they will: 

• pick a winning theme,
• select a fund that is well-placed to harness that theme, and
• see valuations that indicate that the market hasn’t already priced-in the theme’s potential.
• The odds of winning these bets are low, but the prospective payouts can be large.

The long-term performance figures for thematic funds are pretty unflattering. As the chart below shows, just 45 per cent of all thematic funds launched prior to 2010 survived to 2020. Only a fourth managed to both survive and outperform the MSCI World Index over that 10-year span.That said, those that who do win, can win big. For example, those who invested in the ARK Next Generation Internet ETF back at the beginning of 2015 would have raked in annualised returns of 27 per cent through the end of 2019 - three times the return of the MSCI World Index over the same period. It’s performances like these that perpetuate these funds’ allure.

There is no doubt there will be more semiconducter chips in the future, this has been the case for decades and they will be more advanced with each iteration. The industry has always been very cylical though, is the recent increase in prices a strucutral change now? or just a temporary impact of COVID.

I use ETFs for exposure to international markets. Simple on tax, simple to trade via local exchanges.

One of my multiple checks before buying into an ETF is whether the underlying holdings have overlap with any other ETFs I hold.

In the case of SEMI, I compared it to NDQ – which tracks the NASDAQ 100.

Across the top 10 holdings in SEMI, 9 of them are replicated in NDQ.

The 9 holdings: NVIDIA CORP, INTEL CORP, BROADCOM INC, TEXAS INSTRUMENT, QUALCOMM INC, ADV MICRO DEVICE, APPLIED MATERIAL, ANALOG DEVICES, ASML HOLDING NV.

These 9 holdings make up 60.5% of the total weight of SEMI.

These same 9 holdings make up 11.5% of NDQ.

Outcome:

I already have exposure to ~60% of SEMI by holding NDQ. However, NDQ does not offer me exposure to one specific company that make up 9.5% of SEMI, that being TSMC (micro chip maker/ foundry).

I therefore don’t necessarily need to buy SEMI when I already hold NDQ.

I would consider buying SEMI if I developed a thesis where I wanted more weight in the semiconductor industry – but I would need to acknowledge, that both ETFs will likely increase and decrease in concert due to the overlaps.

I would also have to ask as part of that thesis, do I want to hold the SEMI index, or will one or two or three specific winners within the industry or index or thesis outperform overall?

Valuation of an ETF comes down to the valuation of the individual components. Behind the underlying theme are the underlying companies.

These are the top 10 holdings which account for 70% of the Semi ETF:

ASML HOLDINGS NV

NVIDIA CORP

TSMC

INTEL CORP

BROADCOM INC

TEXAS INSTRUMENT

QUALCOMM INC

ADV MICRO DEVICE

APPLIED MATERIAL

ANALOG DEVICES

These companies are all fairly mature businesses, albeit continuing to grow. I'd guess that 15-20%pa CAGR is achievable. That's good, but you aren't going to shoot the lights out with this ETF.